01_First_network.Rmd

title: 'Tutorial: structuration of a semi-distributive GR model'
author: "David Dorchies"
date: "19 mai 2020"
output: html_document

knitr::opts_chunk$set(echo = TRUE)

library(griwrm)

Semi-distributive network description

List of nodes

seine_nodes <- readr::read_delim(
  file = "https://stratus.irstea.fr/d/0b18e688851a45478f7a/files/?p=/climaware_hydro/griwrm_network.txt&dl=1", 
  delim = ";"
)
seine_nodes

Create the ginet object which lists the nodes and describes the network diagram. It's a dataframe of class Ginet and Griwrm with specific column names:

• id: the identifier of the node in the network
• down: the identifier of the next node downstream
• length: hydraulic distance to the next downstream node
• runoff: does the node is a rainfall run-off model?

Ginet function helps to rename the columns of the dataframe and assign the variable classes.

# Specify that all nodes are of run-off type
seine_nodes$runoff <- TRUE
# Convert distance in km as it the unit used by airGR
seine_nodes$length <- seine_nodes$distance_aval / 1000
# Generate the ginet object 
ginet <- Ginet(seine_nodes, list(id = "id_sgl", down = "id_aval"))
ginet

Each line of the Ginet object having the runoff columns switched to TRUE should have a corresponding line in the Girop object which contains the parameters of the rainfall run-off models.

The Girop object is a dataframe of class Girop with specific columns:

• id: the identifier of the node in the network
• area: the total area of the basin (including upstream sub-basins) at the location of the node (km²)
• model: the name of the rainfall run-off model used (e.g. "RunModel_GR4J")
• params: a list containing the calibration parameters of the model

# Specify which run-off model to use
seine_nodes$model = "RunModel_GR4J"
# Generate girop object
girop <- Girop(seine_nodes, list(id = "id_sgl", area = "area"))
girop

Observation time series

Loading hydrometeorological data on the Seine river basin from the ClimAware project:

urls <- 
  file.path(
    "https://stratus.irstea.fr/d/0b18e688851a45478f7a/files/?p=/climaware_hydro/Q_OBS_NAT", 
    paste0(ginet$id, "_NAT.txt&dl=1")
  )
names(urls) <- ginet$id

load_ts <- function(x) {
  ts <- readr::read_delim(file = x, 
             delim = ";", skip = 16, trim_ws = TRUE)
  ts$Date <- as.POSIXlt(lubridate::ymd(ts$Date))
  # Interpolation of data gap in the discharge time serie
  ts[ts$Qnat < 0, "Qnat"] <- NA
  if(is.na(ts$Qnat[nrow(ts)])) {
    ts$Qnat[nrow(ts)] <- 0 # No value at the end: time serie converge to zero
  }
  ts$Qnat <- zoo::na.approx(ts$Qnat)
  ts
}

l <- lapply(urls, load_ts)

Gits object is a list containing a item named date with a timestamp vector of the time series and items named by the identifier of each node. These items contain a dataframe with the observations.

The Gits function creates a Gits object

gits <- Gits(ginet$id[1], l[[ginet$id[1]]], cols = list(date = "Date", Precip = "Ptot", PotEvap = "ETP", Qobs = "Qnat"))

Copy the observations for each node the ginet network:

for(id in ginet$id) {
  l[[id]]$Qnat
  l[[id]]$Qnat <- l[[id]]$Qnat * 86.4 / girop$area[girop$id == id]
  l[[id]]$Qnat[l[[id]]$Qnat < 0] <- NA
  gits <- merge(gits, Gits(id, l[[id]], cols = list(date = "Date", Precip = "Ptot", PotEvap = "ETP", Qobs = "Qnat")))
}

Generate the GRIWRM InputsModel object

The GRIWRM InputsModel object is a list of airGR InputsModel. The identifier of the sub-basin is used as key in the list which is ordered from upstream to downstream.

The airGR CreateInputsModel function is extended in order to handle the ginet object which describe the basin diagram:

InputsModel <- CreateInputsModel(ginet, girop, gits)

Save data for next vignettes

dir.create("_cache", showWarnings = FALSE)
save(ginet, girop, gits, InputsModel, file = "_cache/seine.RData")